Impacts of climate change and inter-annual variability on cereal crops in China from 1980 to 2008

Transcription

1 Research Article Received: 23 April 2 Revised: 26 September 2 Accepted: 23 October 2 Published online in Wiley Online Library: (wileyonlinelibrary.com) DO.2/jsfa.5523 mpacts of climate change and inter-annual variability on cereal crops in China from 98 to 28 Tianyi Zhang a and Yao Huang b Abstract BACKGROUND: Negative climate impacts on crop yield increase pressures on food security in China. n this study, climatic impacts on cereal yields (rice, wheat and maize) were investigated by analyzing climate-yield relationships from 98 to 28. RESULTS: Results indicated that warming was significant, but trends in precipitation and solar radiation were not statistically significant in most of China. n general, maize is particularly sensitive to warming. However, increase in temperature was correlated with both lower and higher yield of rice and wheat, which is inconsistent with the current view that warming results in decline in yields. Of the three cereal crops, further analysis suggested that reduction in yields with higher temperature is accompanied by lower precipitation, which mainly occurred in northern parts of China, suggesting droughts reduced yield due to lack of water resources. Similarly, a positive correlation between temperature and yield can be alternatively explained by the effect of solar radiation, mainly in the southern part of China where water resources are abundant. CONCLUSON: Overall, our study suggests that it is inter-annual variations in precipitation and solar radiation that have driven change in cereal yields in China over the last three decades. c 2 Society of Chemical ndustry Supporting information may be found in the online version of this article. Keywords: China; cereal crops; climate change; climate variability NTRODUCTON China has experienced significant climate change over the last century. For example, the annual mean air temperature increased by. C from 95 to 2. Precipitation in the basins in western China has increased by 5% per decade, while a decrease in precipitation has been observed in northern China. 2 A climatic model has projected that by 25 the annual mean temperature couldriseby Candprecipitationcouldincreaseby5 7%. 2 Additionally, heavy rain could increase along the Yangtze River and in parts of southern, northwest and northeast China. Agricultural land in China accounts for only 7% of the world s arable land area but feeds about 22% of the global population. An increase in crop yields will be required to meet demands of an increasing population. 3,4 Agricultural productivity is dependent upon land resources and climate. Thus, given a finite amount of land, negative impacts of climate change on crop yield will increase stress on food security in China. A number of studies using historical climate and crop data have indicated climatic impacts on cereal yields in Australia, 5 China, 6,7 England, 8 the Philippines, 9 and the USA., A reduction in yields with higher temperature has been found worldwide. 2,3 However, in analyzing observed crop yields over time, it was found that negative correlations between yield and temperature were only predominant for maize, but both positive and negative correlations were detected for rice and wheat cultivated in China. 4 Peng et al. 9 found that an increase in the minimum temperature reduced rice yields on farms at the nternational Rice Research nstitute (RR), but Sheehy et al. 5 pointed out that the yield decrease may be due to reduced radiation, coinciding with a rise in minimum temperature. Several previous studies 6,7 have suggested that covariant climatic variables should be taken into account when evaluating climate yield relationships. However, most current studies are more likely to focus on changes in yield with increasing temperature, which is the most significant feature of climate change. Besides the significant warming trend, a climate model projected that climate change is likely to result in an increase in heavy downpours, while rising air temperature tends to increase evapotranspiration and contribute to dry conditions, 8 suggesting more serious impacts of flood and drought hazards triggered Correspondence to: Tianyi Zhang, State Key Laboratory of Atmospheric Boundary Layer Physics and Atmospheric Chemistry, nstitute of Atmospheric Physics, Chinese Academy of Sciences, Beijing, 29, China. zhangty@mail.iap.ac.cn a State Key Laboratory of Atmospheric Boundary Layer Physics and Atmospheric Chemistry, nstitute of Atmospheric Physics, Chinese Academy of Sciences, Beijing, 29, China b State Key Laboratory of Vegetation and Environmental Change, nstitute of Botany, Chinese Academy of Sciences, Beijing, 93, China J Sci Food Agric (22) c 2 Society of Chemical ndustry

2 T Zhang, Y Huang by climate change. n the past few decades, both of the two hazards have been reported to adversely affect crop production in China. 9,2 Rice, wheat, and maize are the three major cereal crops in China, accounting for approximately 54% of the total sowing area and 89% of the total grain yield. 2 The study examined the correlations between climate and cereal yields (rice, wheat, and maize), effects of climatic variables on cereal yields, and impacts of flood and drought on Chinese agriculture from 98 to 28 on a provincial scale, when the most significant warming trend of the last six decades was detected. 22 MATERALS AND METHODS Study region and crops Rice is grown throughout China. Rice cultivated in the north yields one harvest per year; however, most rice is produced in the south, where double-harvest rice (early and late rice) can be cultivated as well. Wheat and maize are cultivated mostly in the north, while a small portion is produced in the south. Winter wheat is common in most of the study area, while spring wheat is only concentrated in northeast and northwest China. The majority of maize is grown in summer. Harvest area and grain production of cereal crops are illustrated as supporting information Appendix Sa. Our study area is highlighted in Appendix Sb, containing approximately 95% of the total national harvest area. Data sources Crop data were obtained from China s agricultural statistics 2 and the Data Sharing nfrastructure of Earth System Science, 23 extracted from agricultural yearbooks. Data included provincelevel acreage and yield of rice (including early, late, and single rice), wheat (including spring and winter wheat) and maize. Additionally, areas affected by droughts and floods, as well as the total cultivated area from 98 to 28, were also collected from China s agricultural statistics. 2 The crop calendar was derived from the Chinese Agricultural Phenology Atlas. 24 Calendar details are provided in supporting information Appendix S2. Climate data were obtained from the China Meteorological Administration. A total of 48 weather stations are distributed across the study region (supporting information Appendix S3). Climatic variables included daily minimum temperature (T min ), maximum temperature (T max ), sunshine duration and precipitation (Pre). Sunshine duration was converted to solar radiation (Rad) using the Ångström formula. 25 Statistical analysis Correlations between yield and climate Grain yield was correlated with climatic variables using a firstdifference time series 5 (Eqn ()): X t = X t X t () where X represents climatic variables (T min, T max, Pre and Rad) or grain yield (Y). Subscript t denotes the year and includes On a provincial scale, each climatic variable (T min, T max,pre and Rad) was averaged for each crop season in each year. Using Eqn (), T min, T max, Rad, Pre and Y were obtained for each year and province. First-difference values of yield and climatic variables were correlated using a linear regression model for each pair of datasets, including Y with T min, T max, Rad and Pre. Because provincial yields of rice and wheat may include more than one season per year in some provinces (i.e., single, early or late rice, and spring or winter wheat), a weighted average of climatic variables and yields was calculated for each season. Weights were determined by the proportion of cultivated areas in seasons of each year. Correlation between agricultural drought and flood with climate The cultivated area affected by droughts and floods as well as total cultivated area for each province were documented by agricultural statistics, but data on corresponding losses in grain yield due to droughts and floods were not recorded. Therefore, the percentage of areas affected by droughts or floods (drought/floodaffected area divided by total cultivated area) for each province was calculated to represent their impacts on agriculture. A drought indicator (D), calculated as the difference between annual potential evapotranspiration and precipitation (Eqn (2)), was used to represent surface dryness: D = ET p P annual (2) where ET p is the potential evapotranspiration (mm y )calculated by Hargreaves equation; 26 P annual is the annual amount of precipitation (mm y ). To denote annual downpours, a heavy rain index (HR) was calculated, which was defined as total annual rainfall with a rate greater than 5 mm d. 27 Using a linear regression model, the percentage of droughtaffected areas (%drought) was correlated using D and the percentage of flood-affected areas (%flood) was correlated with HR. RESULTS Climate change and variability from 98 to 28 Temperature records from 98 to 28 show a significant warming trend in China, which is consistent with the national assessment report for climate change. 22 Warming was more significant in northern parts of China than in southern parts (Table ). Moreover, in northeast, northern and northwest China, the slopes of T min trend are greater than those of T max in nine provinces of 2 in the regions, whereas, in eastern, central, southwest and southern China, the increases in T min are lower than T max in provinces of 2 in these areas (Table ). Solar radiation from 98 to 28 was not statistically significant in most provinces; however, in five provinces in northern and northwest China, radiation significantly decreased by 5 35 MJ m 2 y per decade (Table ). Annual precipitation in the study region did not show a significant trend. However, a reduction of approximately 3 mm per decade was observed in the Heilongjiang and Sichuan Provinces (Table ). These results suggest that the time trend in radiation and precipitation in most areas of China over the past three decades was not pronounced. Yield correlations with climate Rice n general, increases in the minimum and maximum temperatures did not always reduce rice yield (supporting information Appendix S4). t is noteworthy that rice yield was only significantly reduced as T min increased in Shaanxi Province but it increased with rising wileyonlinelibrary.com/jsfa c 2 Society of Chemical ndustry J Sci Food Agric (22)

3 Climate impact on crops in China Table. Average, standard deviation, and trend in climate over the period in each province T min T max Annual Rad Annual Pre Province D a ( C) Ave. ± SD b Trend c ( Cy ) Ave. ± SD ( C) Trend ( Cy ) Ave. ± SD (MJ m 2 ) Trend (MJ m 2 y ) Ave. ± SD (mm) Trend (mm y ) Northeast Heilongjiang 2.3 ± ± ± ± Jilin 2.2 ± ± ± ± 74 2 Liaoning ± ± ± ± 4 3 North Beijing 4 8. ± ± ± ± Hebei ± ± ± ± 86 2 Tianjin 6 9. ± ± ± ± 23 6 Shandong 7 8. ± ± ± ± Henan 8. ± ± ± ± 35 4 Northwest Shanxi 9 3. ± ± ± ± 7 8 Shaanxi 7. ± ± ± ± 9 46 Ningxia 2.6 ± ± ± ± 53 Gansu 2.3 ± ± ± ± 34 4 East Anhui 3.4 ± ± ± ± 7 7 Jiangsu 4.8 ± ± ± ± 58 5 Zhejiang 5 4. ± ± ± ± 8 25 Central Hubei ± ± ± ± Hunan ± ± ± ± 67 3 Jiangxi ± ± ± ± Southwest Guizhou ± ± ± ± 6 Sichuan ± ± ± ± 67 3 Yunnan 2.9 ± ± ± ± 88 2 South Fujian ± ± ± ± 29 4 Guangdong ± ± ± ± Guangxi ± ± ± ± a Same as in Appendix Sb. b Averaged over the period SD, standard deviation. c Slope of the linear regression of climate variable with time in years. Asterisks indicate significance at confidence levels of 5% and %, respectively. T min in Heilongjiang, Yunnan, and Guangxi Province (Fig. (a)). An increase in T max significantly decreased rice yield in four provinces and improved yield in two provinces (Fig. (b)). Yield was significantly correlated with Rad (Fig. (c)) and Pre (Fig. (d)) in four of the 24 provinces (Appendix S4). mportantly, there is a clear spatial distribution for correlation between yield and Pre. The positive correlations mainly occurred in the provinces of northern and northwest China as well as Sichuan and Guizhou Provinces in southwest China, even through these correlations are insignificant. For other regions, yields became lower with higher precipitation (Fig. (d)). Correlations between yield and climatic variables (Appendix S4 for rice) were plotted on a province-by-province basis (Fig. 2). t is observed that when the correlation between yield and temperature ( T min and T max ) was positive, yields were also positively correlated with Rad (quadrant in Fig. 2(a,b)) and negatively correlated with Pre (quadrant V in Fig. 2(c,d)). Conversely, for the data points with negative correlation between yield and temperature, a negative correlation between yield and Rad (quadrant in Fig. 2(a,b)) and a positive correlation between yield and Pre (quadrant in Fig. 2(c,d)) tended to occur simultaneously. These results indicated that correlations between yield and temperature could be alternatively explained by effects of radiation and precipitation. Wheat n general, the correlation between wheat yield and T min was not statistically significant (Fig. 3(a)); however, an increase in T max appeared to reduce wheat yield in northeast China (Fig. 3(b)). Wheat yield in three provinces in eastern China and one province in southern China was significantly correlated with radiation in a positive manner (Fig. 3(c)). n contrast, an increase in precipitation generally reduced wheat yield in eastern, central and southern China (Fig. 3(d)). Figure 4 plots the correlation coefficients between wheat yield and climatic variables (Appendix S4 for wheat) in one graph. J Sci Food Agric (22) c 2 Society of Chemical ndustry wileyonlinelibrary.com/jsfa

4 T Zhang, Y Huang (a) (b) (c) (d) Figure. Spatial distribution of correlation coefficient between yield and climate for rice: (a) Y vs. T min ;(b) Y vs. T max ;(c) Y vs. Rad; (d) Y vs. Pre. Statistical significance at P <.5. See Appendix S4 for details. There is a positive correlation between wheat yield and T max when yields were also positively correlated with Rad (quadrant in Fig. 4(b)) and negatively correlated with Pre (quadrant V in Fig. 4(d)), and vice versa (data points fell into quadrant in Fig. 4(b) and into quadrant in Fig. 4(d), respectively). Such covariant climatic effects are unclear, however, for T min (Fig. 4(a,c)). Maize Maize yield was significantly reduced (Fig. 5(a,b)) in approximately 5% of provinces (Appendix S4) with an increase in temperature. Maize yield was more negatively correlated with T max than T min (Fig. 5(a,b)). Additionally, yield was negatively correlated with radiation (Fig. 5(c)) and positively correlated with precipitation (Fig. 5(d)) in most areas, with six and seven provinces, respectively, showing a significant relationship (Appendix S4). A negative correlation between maize yield and T min or T max was coincident with a negative correlation between yield and Rad (quadrant in Fig. 6(a,b)) and a positive correlation between yield and Pre (quadrant in Fig. 6(c,d)). These results suggest that a lower maize yield is accompanied by a decrease in precipitation, an increase in radiation or an increase in temperature. Correlations between climatic variables Correlations between climatic variables are quite significant in China (Appendix S4). For the growing season of the three crops, on an inter-annual basis, higher T max is accompanied by higher Rad and lower Pre in a significant manner in most areas, while the correlations for T min are relative weak. Rad is significantly correlated with Pre in a negative manner in most areas. Percentage of cultivated area significantly impacted by climate The percentage of cultivated area significantly impacted by climate was estimated by correlations between yield and climatic variables (AppendixS4)andthecultivatedarea(themeanfrom98to28) of rice, wheat, and maize in each province. An increase in minimum temperature did not lead to a significant reduction in rice and wheat yields; in particular, rice yield increased in 4.8% of the sowing areas due to an increase in T min (Table 2). An increase in the maximum temperature caused a lower yield in 2.7% of the rice-growing areas and.3% of the wheat-growing areas (Table 2). Maize was more sensitive to warming than rice and wheat: 6.2% and 35.5% of maize-cultivated areas showed significant negative correlations with T min and T max, respectively (Table 2). Precipitation had a stronger impact on yield than temperature. Lower yield with higher precipitation occurred in 32.5% of rice and 8.4% of wheat but higher precipitation increased maize yield in 44.4% of the cultivated area (Table 2). The reduction in rice and wheat yields occurred mainly in central and southern China, where annual precipitation was greater than mm y (Table ). Positive effects of precipitation on maize yield appeared in the northern part of China (Fig. 5(d)), where annual precipitation is generally lower than 7 mm (Table ). Thus maize yield in these provinces was more vulnerable to precipitation than to temperature. wileyonlinelibrary.com/jsfa c 2 Society of Chemical ndustry J Sci Food Agric (22)

5 Climate impact on crops in China Y vs. Rad (r) Y vs. Pre (r).8 (a) V (c) V Y vs. T min (r).8 (b) V (d) V Y vs. T max (r) Figure 2. Pattern of rice yield correlations to climatic variables. Values show the Pearson correlation coefficient between rice yield and climate variables.,,, V denote the quadrant. Table 2. Percentage of cultivated area (%, mean from 98 to 28) with insignificant, significant positive and negative (P <.5) impact of climatic parameter on yield Climatic variable Correlation with Y Rice Wheat Maize T min nsignificant Positive Negative T max nsignificant Positive Negative Pre nsignificant Positive Negative Rad nsignificant Positive Negative An increase in radiation reduced maize yield in 4.6% of the cultivated area (Table 2). Although this appears counterintuitive, radiation was negatively correlated with precipitation (Appendix S4); thus lower annual precipitation was accompanied by higher solar radiation. The resultant water stress conditions caused a reduction in yield (Appendix S4). mpact of droughts and floods At a regional level, droughts affected a larger cropland area than floods, with an average of 22.5 ± 6.9 Mhaand.5 ± 4.8 Mha, respectively, affected by drought, which encompasses approximately 6 ± 5% and 8 ± 3% of the total area (Table 3). n the northern part of China, upland crops were dominant and drought was more significant than in other regions of China. Areas affected by drought accounted for 24 ± 4% of the total cultivated area. For floods, impacts were stronger in northeast, central and southeast provinces, where a rotation of rice with upland crops or double rice systems was common. Areas affected by floods accounted for ± % of the total cultivated area in the study region (Table 3). No significant trends in the percentage of drought- and floodaffected areas were observed from 98 to 28 at the national level (Table 3). At the provincial level, percentage of droughtaffected areas decreased in Hebei, Shandong, Henan, and Hunan Province, while the percentage of flood-affected areas decreased in Heilongjiang Province over the period (Table 3). n general, D tended to increase over time throughout the region. However, it was only statistically significant in three provinces (Heilongjiang, Shaanxi, and Sichuan). HR decreased in Beijing and increased in Fujian, but no significant trends were detected in other provinces (Table 3). The percentage of areas affected by drought (%drought) correlated significantly with D (Table 3). An increase in evapotranspiration or a decrease in precipitation (Eqn (2)) tended to enhance drought stress, with the largest impacts in northeast and northwest regions. According to correlations between %drought and D, a mm increase in D caused an additional 4 5% of the total cultivated area to be affected by drought in the regions. However, a mm increase in the D of eastern, central, southwest J Sci Food Agric (22) c 2 Society of Chemical ndustry wileyonlinelibrary.com/jsfa

6 T Zhang, Y Huang (a) (b) (c) (d) Figure 3. Spatial distribution of correlation coefficient between yield and climate for wheat: (a) Y vs. T min ;(b) Y vs. T max ;(c) Y vs. Rad; (d) Y vs. Pre. Statistical significance at P <.5. See Appendix S4 for details. and southern provinces only led to an increase in drought-affected area by 4% (Table 3). The percentage of areas affected by floods (%flood) was positively correlated with HR. Crops in northeast China were most sensitive to heavy downpours: a mm y increase in heavy rain affected an additional 5 43% of the cultivated area (Table 3). Although annual precipitation was quite high, the crop area in the south was less sensitive to heavy downpours (Table ). The above results give us a general picture of drought and flood impacts in China. n northern parts, drought impacts are much stronger than floods; in southern parts, drought impacts are weaker but floods show no significantly higher effect than droughts. DSCUSSON t is often thought that an increase in temperature causes a decrease in the yields of Chinese cereal crops. 6,28 However, data obtained in the present study revealed that an increase in temperature had no significant effect on cereal yields at a national scale and were associated with both an increase and decrease in yields from 98 to 28. Of the three crops in this investigation, maize seemed most vulnerable to higher temperatures, while rice and wheat were not so sensitive (Table 2). Thus the often cited result that warming has a negative impact on yield does not explain the observations of this study. Correlations between yield and temperature obtained by this study are consistent with the results of Tao et al. 4 However, our analysis suggests that the varying correlation of yield with temperature is due to covariant effects of other climatic variables intertwined with temperature. During the growing season, a higher temperature was generally accompanied by less precipitation in the majority of areas (Appendix S4). Consequently, a negative correlation between yield and temperature can be interpreted as a positive correlation with precipitation, mainly in northern provinces (Figs 2, 4, and 6), especially for maize. This result indicates that drought significantly limits yield. 29 Moreover, data obtained in this study revealed that most drought-affected areas were located in the northern part of China (Table 3), which is consistent with above climate yield correlations. Significant impacts on yield due to drought have been reported in other regional scale assessments. Lobell and Burke 3 found that crop yields in most countries were negatively correlated with temperature but they were also positively correlated with precipitation in a more significant manner. This reflects that precipitation has a more significant impact on crop yield than temperature. n Australia, Nicholls 5 found a negative correlation between wheat yield and diurnal temperature range. However, Gifford et al. 3 indicated that this negative correlation may be due to droughts. n China, significant droughts in north have been widely recognized (Table 3) and are clearly due to a lack of water resources For example, droughts highly constrained the production of rice, 36,37 wheat, 38,39 and maize 4,4 in the region. Warmer temperatures also coincided with higher levels of solar radiation (Appendix S4). mproved photosynthesis associated with an increase in radiation has a positive effect on yield, 42 which could explain an increase in yield with higher temperatures (Figs 2, 4, and 6). Radiation driving crop yield is not new. For instance, Peng wileyonlinelibrary.com/jsfa c 2 Society of Chemical ndustry J Sci Food Agric (22)

7 Climate impact on crops in China Y vs. Rad (r).8 (a) V (b) V Y vs. Pre (r).8 (c) V Y vs. T min (r) (d) Y vs. T max (r) V Figure 4. Pattern of wheat yield correlations to climatic variables. Values show the Pearson correlation coefficient between wheat yield and climate variables.,,, V denote the quadrant. et al. 9 suggested a strong negative minimum temperature effect on rice yield from long-term experiments at RR that maintained good growing conditions for crop development. However, Sheehy et al. 5 pointed out that the observed reduction in yield at RR in Peng s study could be largely explained by a significant negative correlation between T min and Rad; thus an effect due exclusively to minimum temperature was not necessary to explain observed trends in yield, and radiation is largely responsible for change in yield. n our study, there are seldom significant negative correlations between T min and Rad in most provinces of China (Appendix S4), which is in contrast to the aforementioned situation at RR and validates Sheehy s hypothesis. n China, recent studies based on long-term field experiments also found that radiation has a stronger correlation with yield than temperature, 43 which is in accordance with the effects of radiation on yield when positive correlations between yield and temperature were presented (Figs 2, 4, and 6). The above analysis reveals that changes in precipitation and solar radiation drove variations in yields of cereal crops. Yields of rice, wheat, and maize were sensitive to drought in the northern part of China. Maize was most sensitive to droughts (Fig. 5(d)), while rice was least sensitive (Fig. (d)), which can be explained by rice having nearly twice as large an irrigated area (95%) as maize (45%) allocated in China. 44 n southern parts, however, yields tended to decrease with greater amounts of precipitation (negative correlation), which is especially evident in rice (Fig. (d)) and wheat (Fig. 3(d)). The decrease in yield with higher precipitation is not due to floods since impacts of floods were shown to be not very strong in the south relative to drought (Table 3). This is probably due to lower radiation accompanied by higher precipitation (Appendix S4). n summary, variations in yield related to climate showed a strong inter-annual variability, where changes in precipitation and radiation were significant (Table ), even though an increase in temperature is the most significant feature of climate change. CONCLUSONS n China, a significant warming trend was observed from98 to 28, but trends in annual precipitation and solar radiation were not statistically significant for a majority of the region. n general, maize was relatively sensitive to increasing temperatures, while an increase in temperature was correlated not only with lower but also with higher yields of rice and wheat. The observation suggests that the current view that warming adversely affects cereal yield cannot be held to be universally true. Further investigation found that negative correlation between yields and temperature tended to be coincident with positive yield correlations with precipitation, mainly in the northern part of China where water resources are limiting and droughts have affected a large area. On the other hand, a positive correlation between yield and temperature was generally accompanied by positive yield correlations with radiation, primarily in southern parts where much richer water resources meet the water requirements of crops but flood impacts are not strong. This reflects that effects of precipitation (via drought) and radiation (via photosynthetic processes) intertwined with temperature variation explain change in yields of the three crops in China, suggesting a covariant impact of climatic variables. Moreover, no significant trends in the percentage of drought- and flood-affected areas were observed over the period. n summary, J Sci Food Agric (22) c 2 Society of Chemical ndustry wileyonlinelibrary.com/jsfa

8 T Zhang, Y Huang (a) (b) (c) (d) Figure 5. Spatial distribution of correlation coefficient between yield and climate for maize: (a) Y vs. T min ;(b) Y vs. Tmax; (c) Y vs. Rad; (d) Y vs. Pre. Statistical significance at P <.5. See Appendix S4 for details. Y vs. Rad (r) Y vs. Pre (r) (a) (c) V Y vs. T min (r) V (b) (d) Y vs. T max (r) V V Figure 6. Pattern of maize yield correlations to climatic variables. Values show the Pearson correlation coefficient between maize yield and climate variables.,,, V denote the quadrant. wileyonlinelibrary.com/jsfa c 2 Society of Chemical ndustry J Sci Food Agric (22)

9 Climate impact on crops in China Table 3. Percentage of drought- and flood-affected areas, trend and correlation with D and HR % drought % drought vs. D d % flood % flood vs. HR d Region and province D a Ave. ± SD b (%) Trend c (% y ) D Trend (mm y ) Slope (% mm ) r Ave. ± SD (%) Trend (% y ) HR Trend (mm y ) Slope (% mm ) r Northeast Heilongjiang 2 ± ± Jilin 2 26 ± ± Liaoning 3 29 ± ± North Beijing 4 6 ± ± Hebei 5 2 ± ± Tianjin 6 22 ± ± Shandong 7 2 ± ± Henan 8 6 ± ± Northwest Shanxi 9 38 ± ± Shaanxi 27 ± ± Ningxia 22 ± ± Gansu 2 29 ± ± East Anhui 3 2 ± ± Jiangsu 4 ± ± Zhejiang 5 7 ± ± Central Hubei 6 3 ± ± Hunan 7 3 ± ± Jiangxi 8 8 ± ± Southwest Guizhou 9 7 ± ± Sichuan 2 2 ± ± Yunnan 2 2 ± ± South Fujian 22 7 ± ± Guangdong 23 8 ± ± Guangxi 24 2 ± ± National total 6 ± ± 3.3 a Same as in Appendix Sb. b Averaged over the period SD, standard deviation. c Slope of the linear regression with time in years. d A linear regression of % droughts with D, or % floods with HR. Significant at confidence levels of 5% and %, respectively. this study revealed that inter-annual variation in precipitation and solar radiation are the major climatic drivers of the three cereal crop yields in China over the last three decades. Supporting information Supporting information may be found in the online version of this article. ACKNOWLEDGEMENTS This research was jointly supported by the External Cooperation Program of the Chinese Academy of Sciences (Grant No. GJHZ24), China Meteorological Administration (Grant No ) and the Ministry of Finance People s Republic of China (Grant No. GYHY2868). We also thank Data Sharing nfrastructure of Earth System Science and China Meteorological Administration for providing agricultural and climate data, and anonymous reviewers for insightful suggestions. REFERENCES Ding Y, Ren G, Zhao Z, Xu Y, Luo Y, Li Q, et al, Detection, causes and projection of climate change over China: an overview of recent progress. Adv Atmos Sci 24: (27). 2 Qin D, Ding Y, Su J, Ren J, Wang S, Wu R, et al, Assessment of climate and environment changes in China. : Climate and environment changes in China and their projections. Adv Climate Change Res 2: 5 (26). 3 Evans LT, Feeding the Ten Billion: Plants and Population Growth. Cambridge University Press, Cambridge, UK (998). 4 Greenland DJ, World food security: perspectives past, present, and future, in Climate Change and Global Food Security, ed. by Lal R, Uphoff N, Stewart BA and Hansen DO. CRC press, London, pp 2 38 (25). J Sci Food Agric (22) c 2 Society of Chemical ndustry wileyonlinelibrary.com/jsfa

10 T Zhang, Y Huang 5 Nicholls N, ncreased Australian wheat yield due to recent climate trends. Nature 387: (997). 6 Tao F, Yokozawa M, Xu Y, Hayashi Y and Zhang Z, Climate changes and trends in phenology and yields of field crops in China, Agric Forest Meteorol 38:82 92 (26). 7 You L, Rosegrant MW, Wood S and Sun D, mpact of growing season temperature on wheat productivity in China. Agric Forest Meteorol 49:9 4 (29). 8 Chmielewski FM and Potts JM, The relationship between crop yields from an experiment in southern England and long-term climate variations. Agric Forest Meteorol 73:43 66 (995). 9 Peng S, Huang J, Sheehy JE, Laza RC, Visperas RM, Zhong X, et al,rice yields decline with higher night temperature from global warming. Proc Natl Acad Sci USA : (24). Lobell DB and Asner G, Climate and management contributions to recent trends in U.S. agricultural yields. Science 299:32 (23). Schlenker W and Roberts MJ, Nonlinear temperature effects indicate severe damages to U.S. crop yields under climate change. Proc Natl Acad Sci USA 6: (29). 2 Lobell DB, Burke MB, Tebaldi C, Mastrandrea MD, Falcon WP and Naylor RL, Prioritizing climate change adaptation needs for food security in 23. Science 39:67 6 (28). 3 Lobell DB, Changes in diurnal temperature range and national cereal yields. Agric Forest Meteorol 45: (27). 4 Tao F, Yokozawa M, Liu J and Zhang Z, Climate crop yield relationships at provincial scales in China and the impacts of recent climatetrends. Climate Res 38:83 94 (28). 5 Sheehy JE, Mitchell PL and Ferrer AB, Decline in rice grain yields with temperature: models and correlations can give different estimates. Field Crop Res 98:5 56 (26). 6 Zhang T, Zhu J and Wassmann R, Responses of rice yields to recent climate change in China: an empirical assessment based on longterm observations at different spatial scales (98 25). Agric Forest Meteorol 5:28 37 (2). 7 Lobell DB and Oritiz-Monasterio J, mpacts of day versus night temperature on spring wheat yields: a comparison of empirical and CERES model predictions in three locations. Agron J 99: (27). 8 Karl TR, Melillo JM and Peterson TC (eds), Global Climate Change mpactsintheunitedstates. Cambridge University Press, Cambridge, UK (29). 9 Feng Q, Chang G and Mikami M, Water resources in China: problems andcountermeasures. AMBO 28:22 23 (999). 2 Tao F, Yokozawa M, Hayashi Y and Lin E, Changes in agricultural water demands and soil moisture in China over the last half-century and their effects on agricultural production. Agric Forest Meteorol 8:25 26 (23). 2 Editorial board, Agricultural Statistics of New China in Sixty Years. China Agricultural Press, Beijing (in Chinese) (29). 22 Edition board, National Assessment Report for Climate Change in China. Science Press, Beijing (in Chinese with English abstract) (27). 23 Data Sharing nfrastructure of Earth System Science, nstitute of Geographical Sciences and Natural Resources Research, Chinese Academy of Sciences (29). [Online]. Available: November 2]. 24 Zhang F, Chinese Agricultural Phenology Atlas.SciencePress,Beijing (in Chinese) (987). 25 Almorow J and Hontoria C, Global solar radiation estimation using sunshine duration in Spain. Energy Convers Manage. 45: (24). 26 Hargreaves GH and Allen RG, History and evaluation of Hargreaves evapotranspirationequation. J rrig Drain E-ASCE 29:53 63 (23). 27 Gao X and Zhao Z, Changes of extreme events in regional climate simulationsovereastasia. Adv Atmos Sci 9: (22). 28 Wang HL, Gan YT, Wang RY, Niu JY, Zhao H, Yang QG, et al, Phenological trends in winter wheat and spring cotton in response to climate changes in northwest China. Agric Forest Meteorol 48: (28). 29 Wopereis MC, Kropff MJ, Maligaya AR and Tuong TP, Drought-stress response of two lowland rice cultivars to soil and water status. Field Crop Res 46:2 39 (996). 3 Lobell DB and Burke M, Why are agricultural impacts of climate change so uncertain? The important of temperature relative to precipitation. Environ Res Lett 3:347 (28). 3 Gifford R, Angus J, Barrett D, Passioura J, Rawson H, Richards R, et al, Comments on Climate change and Australian wheat yield. Nature 39:448 (998). 32 Changming L and Jingjie Y, Groundwater exploitation and its impact on the environment in the North China Plain. Water nt 26: (2). 33 Shu G, Yixing Z, Minghua Z and Smallwood KS, A sustainable agroecological solution to water shortage in the North China Plain (HuabeiPlain). J Environ Manage 44: (2). 34 Khan S., Hanjra MA and Mu J, Water management and crop production for food security in China: a review. AgricWaterManage 96: (29). 35 Varis O and Vakkilainen P, China s 8 challenges to water resources management in the first quarter of the 2st century. Geomorphology 4:93 4 (2). 36 Wang H, Bouman BAM, Zhao D, Wang C and Moya PF, Aerobic rice in northern China: opportunities and challenges, in Water- Wise Rice Production, ed. by Bouman BAM, Hengsdijk H, Hardy B, Bindraban PS, Tuong TP and Ladha JK. Proceedings of the nternational Workshop on Water-Wise Rice Production, 8 April, nternational Rice Research nstitute, Los Baños, Philippines, pp (22). 37 Bouman BAM, Yang X, Wang H, Wang Z, Zhao J and Chen B, Performance of aerobic rice varieties under irrigated conditions in North China. Field Crop Res 97:53 65 (25). 38 Kang S, Zhang L, Liang Y, Hu X, Cai H and Gu B, Effects of limited irrigation on yield and water use efficiency of winter wheat in the Loess Plateau of China. Agric Water Manage 55:23 26 (22). 39 Zhang H, Wang X, You M and Liu C, Water-yield relations and wateruse efficiency of winter wheat in the North China Plain. rrigation Sci 9:37 45 (999). 4 Zhang Y, Kendy E, Qiang Y, Changming L, Yanjun S and Hongyong S, Effect of soil water deficit on evapotranspiration, crop yield, and water use efficiency in the North China Plain. Agric Water Manage 64:7 22 (24). 4 Zhang X, Pei D. Chen S, Sun H and Yang Y, Performance of doublecropped winter wheat summer maize under minimum irrigation in the North China Plain. Agron J 98: (26). 42 Stanhill G and Cohen S, Global dimming: a review of the evidence for a widespread and significant reduction in global radiation with discussion of its probable causes and possible agricultural consequences. Agric Forest Meteorol 7: (2). 43 Zhang T, Zhu J, Yang X and Zhang X, Correlation changes between rice yields in North and Northwest China and ENSO from 96 to 24. Agric Forest Meteorol 48:2 33 (28). 44 Huang Q, Rozelle S, Lohmar B, Huang J and Wang J, rrigation, agricultural performance and poverty reduction in China. Food Policy 3:3 52 (26). wileyonlinelibrary.com/jsfa c 2 Society of Chemical ndustry J Sci Food Agric (22)